基于BP网络的齿轮弯曲疲劳强度极限的确定

Abstract

The bending fatigue limits is generally given from a chart when designing the strength of the gear,which has brought great inconvenience to driving the design of CAD with many charts.Considering the impact factors of the bending fatigue limit（gear materials,hardness of heat treatment,materials and heat treatment quality） and the nonlinear features between these impact factors and gear bending fatigue limit,A method of using four-storey BP neural network model to determine the gear bending fatigue limit is put forward to.An analytical expression is established based on two special points of bending strength limit,using its value as BP network training samples.The trained BP network model can be directly used for establishing the limits of gear bending fatigue strength in engineering design,some cases confirmed the error of no more than 5 percent.